AIM_COACH is a proof-of-concept study investigating Brain-Computer Interface (BCI)-based cognitive monitoring using High-Density Electroencephalography (HD EEG) in infants. Its main goal is to assess the feasibility of providing BCI-based cognitive monitoring within an infant's first year.
Infancy is a crucial time for sensory development, especially in shaping how infants perceive their environment. This study focuses on visual motion perception and analyzes EEG recordings related to looming objects. It involves 100 full-term and 26 preterm infants at two developmental stages: 3-5 months, the pre-locomotor phase and 10-12 months, the crawling phase.
During the experiments, infants were exposed to a visual stimulus simulating an object approaching rapidly, designed to study their neural responses to potential threats (refer to Figure 1). All participating infants were healthy, with no reported complications, and informed consent was obtained from their parents or guardians. Ethical approval for the study was granted by relevant committees.
Overview of results: AIM_COACH study on looming-related brain responses in infants has unveiled shared patterns and subtle age-related variations during impending collision events. These neural patterns are consistent across all infants (full-term and preterm), primarily concentrated in the parieto-occipital cortices. Both groups of infants exhibited similar responses to looming stimuli, characterized by theta and alpha frequency bands' phase-locking and theta band event-related synchronization.
To identify looming-related brain responses, a Support Vector Machine (SVM) classifier was trained using data from older full-term infants. This classifier effectively identified such responses in younger full-term infants without significant performance decline. Specifically, the classifier trained on amplitude and power features achieved above-chance accuracy, while phase features did not perform significantly better than chance.
The classification performance of these features on younger infants was evaluated, and amplitude yielded the highest accuracy, suggesting the viability of real-time detection of looming responses in infants through EEG and BCI technology. This supports the future development of BCI for monitoring infant perceptual development in real-time.
As infants grow older and more mobile, improvements in looming-related brain responses become evident. Older infants displayed responses closer to the virtual collision point and greater activity in higher alpha-band frequencies. Notably, full-term infants exhibited reduced latencies, especially in time-to-contact (ttc), as they matured, while preterm infants did not show similar improvements. The variation in ttc values suggests the potential use of latency as a developmental biomarker within the BCI paradigm. This marker could assess infants' visual maturity, offering insights into their perceptual growth.
AIM_COACH has shared its findings through journal publications, including articles submitted to IEEE Transactions on Cognitive and Developmental Systems (under review) and Brain Sciences (under preparation). The project actively participated in conferences and workshops, such as the International BCI Meeting and International Congress of Infant Studies, to present research outcomes and engage with experts in the field. AIM_COACH maintains a dedicated website featuring articles summarizing its research findings and related studies aimed at informing parents and caregivers.
